Captured retina images enable important parts of the visual system to be analyzed. Automated retinal image processing is becoming a primary screening tool for the detection of diseases such as diabetic retinopathy (DR). An automated system reduces human error and also reduces the burden on ophthalmologists. The accurate detection of microaneurysms (MAs) is an important step for the early detection of DR. MAs appear as a first sign of DR and can be seen on retina images. This paper discusses some of the current techniques used to automatically detect MAs from retinal digital fundus images. This review outlines the general principle upon which retinal digital image analysis is based for the detection of MAs. The algorithms are categorized according to four processing steps (preprocessing, candidate MA detection, feature extraction, and classification). Various gold standard or ground truth databases, data sample size, and the use of image databases are discussed. The variety of outcome measures and flaws in the literature are discussed. The challenges and future potential for research are discussed to provide guidance to algorithm designers of the early detection of DR.